Module markdown 

Markdown -> Product extraction (PORT-CONTRACT.d/03, parity landmine #2).

Bespoke CommonMark block-boundary tokenizer reproducing markdown-it-py 4.2.0’s "commonmark" preset for the consumed surface only: heading_open (tag + 0-based start line) and inline (raw content + start line), plus the oracle’s section-slicing walk on top (src/rac/core/markdown.py).

The bake-off of PORT-CONTRACT decision 4 resolves to bespoke: external markdown crates are not in the dependency allowlist and the consumed surface is block boundaries only, so the block rules of markdown-it-py (code fence blockquote hr list reference html_block heading lheading paragraph) are ported line-by-line from the oracle venv’s source.

Python-semantics helpers (py_strip, py_casefold, py_repr_str, Unicode \d) come from crate::pycompat per the layer-1 contract.

Structs

BlockEvent

One consumed-surface event: a heading (tag + raw inline text) or a body inline (raw content), each with the block’s 0-based start line.

FrontmatterSplit

split_frontmatter result (decided.core.frontmatter.FrontmatterSplit).

Issue

MalformedRequirement

Product

Requirement

SearchSection

Token

Constants

DEFAULT_MAX_FILE_BYTES

MAX_CAPTURED_LINES

MAX_FIELD_CHARS

Functions

consumed_events

The exact token surface the oracle walk consumes (contract section 1).

max_file_bytes

The per-file byte cap, honoring DECIDED_MAX_FILE_BYTES (REQ-001).

max_file_bytes_from

max_file_bytes() computed from a raw env value (None = unset).

parse

Parse Markdown text into a Product (decided.core.markdown.parse), with the byte cap taken from the environment.

parse_file

Read path and parse it (decided.core.markdown.parse_file).

parse_file_with_cap

parse_file with an explicit byte cap (testing seam).

parse_with_cap

parse with an explicit byte cap (testing seam; the oracle reads the env).

py_parse_int

Python int(str) (decimal): strips the Unicode whitespace set (which, unlike str.strip(), excludes U+001C-U+001F — verified empirically, and it matches Rust str::trim), accepts an optional sign, Unicode Nd digits, and single underscores strictly between digits.

split_frontmatter

Split a leading --- frontmatter block (decided.core.frontmatter).

tokenize_blocks

Tokenize a (frontmatter-stripped) Markdown body into the flat block-level token stream, exactly as markdown-it-py 4.2.0 “commonmark” emits it.